The idea of explicitly associating a measure of "value durability" with electronic artifacts was invented at Scrutable Systems in the late 1990s, as we began recognizing inherent business advantages of XML representations over most other artifacts. (Here, an "artifact" is any electronic media resource or file). To understand how this idea applies to ontology knowledge capture, we should consider a few points of comparison between ontology mechanisms and some other mechanisms widely used in software engineering today.
A bit of web research will reveal that OWL is currently the leading standard for capturing ontologies. Very briefly, an ontology represented in OWL is a specification utilizing several vocabulary elements:
- Classes of entities (similar to classes in a UML object model, or tables in a relational database)
- Properties of classes (similar to the member variables of a class, or columns in a relational database)
- Individuals belonging to classes, and their relationships to other individuals (similar to instances in a running OO program, or rows in a relational database. Note that UML is not very well suited to representing instance data)
- Constraints on classes and properties (similar to OCL constraints in a UML model, or integrity constraints in a relational database)
The precise OWL definitions of these concepts are very interesting, if somewhat subtle, but let's ignore them for the moment. Instead, let's point out a few more rather well known things about OWL models:
- OWL models are RDF models, and RDF models can be encoded as XML. (We emphasize "can" because RDF models can also be encoded as SQL or in other ways). Since it is technically easy to share XML documents between software applications, regardless of the programming language and operating system platform, it is also easy to share RDF models and therefore OWL models.
- OWL and RDF models may fully exploit the capacity of URIs and XML-namespaces to describe information resources unambiguously. Morover, techniques exist for representing multiple OWL models within a single XML document, for representing a single model spread over multiple documents, etc.
- The conceptual definitions in these models may be refined over time without requiring that all instance data be wiped out and recreated.
- OWL models do not represent "behavior", per se. They are entirely declarative.
Taken together, we claim that these observations imply the following statements about OWL models in comparison with SQL, UML, Java or C# artifacts.
- Because it is designed for distributed information description rather than program definition, OWL is more powerful for describing real-world concepts than UML or Java/C#.
- Because it explicitly allows for inheritance and many other subtle
relationships, OWL models are also more powerful than SQL schemas.
- Because they eschew "behavior", OWL models may be edited directly by
non-programmers, i.e. domain experts, using appropriate tools.
- Because they are XML based, OWL models may be easily manipulated and exchanged by applications, regardless of platform. Furthermore, it is intrinsically tractable to extract the contents of an OWL model into any other format, including those not yet invented.
These are broad generalizations, and so we might find experts who would take issue with some aspects of these statements. However, if we accept them for a moment at face value, then what are the implications of these statements for business today?
At LogicU, we think the most important implication is this: for purposes of capturing conceptual relationships, artifacts created today using OWL/RDF are better positioned to maintain their value over the coming years than are UML models, object oriented source files, or SQL schemas. Together with the increased expressive power offered by the ontology approach, this durability means we can expect a better long-term return from an investment in ontology artifacts than from a comparable investment in the alternatives.
Now that we've explained the relevance and advantages of OWL ontologies, you are invited to read about The LogicU Curriculum.